China's Rush To Dominate AI Comes With a Twist: It Depends on US Technology (nytimes.com) 32
China's tech firms were caught off guard by breakthroughs in generative artificial intelligence. Beijing's regulations and a sagging economy aren't helping. From a report: In November, a year after ChatGPT's release, a relatively unknown Chinese start-up leaped to the top of a leaderboard that judged the abilities of open-source artificial intelligence systems. The Chinese firm, 01.AI, was only eight months old but had deep-pocketed backers and a $1 billion valuation and was founded by a well-known investor and technologist, Kai-Fu Lee. In interviews, Mr. Lee presented his A.I. system as an alternative to options like Meta's generative A.I. model, called LLaMA. There was just one twist: Some of the technology in 01.AI's system came from LLaMA. Mr. Lee's start-up then built on Meta's technology, training its system with new data to make it more powerful.
The situation is emblematic of a reality that many in China openly admit. Even as the country races to build generative A.I., Chinese companies are relying almost entirely on underlying systems from the United States. China now lags the United States in generative A.I. by at least a year and may be falling further behind, according to more than a dozen tech industry insiders and leading engineers, setting the stage for a new phase in the cutthroat technological competition between the two nations that some have likened to a cold war. "Chinese companies are under tremendous pressure to keep abreast of U.S. innovations," said Chris Nicholson, an investor with the venture capital firm Page One Ventures who focuses on A.I. technologies. The release of ChatGPT was "yet another Sputnik moment that China felt it had to respond to."
Jenny Xiao, a partner at Leonis Capital, an investment firm that focuses on A.I.-powered companies, said the A.I. models that Chinese companies build from scratch "aren't very good," leading to many Chinese firms often using "fine-tuned versions of Western models." She estimated China was two to three years behind the United States in generative A.I. developments. The jockeying for A.I. primacy has huge implications. Breakthroughs in generative A.I. could tip the global technological balance of power, increasing people's productivity, aiding industries and leading to future innovations, even as nations struggle with the technology's risks. As Chinese firms aim to catch up by turning to open-source A.I. models from the United States, Washington is in a difficult spot. Even as the United States has tried to slow China's advancements by limiting the sale of microchips and curbing investments, it has not held back the practice of openly releasing software to encourage its adoption. For China, the newfound reliance on A.I. systems from the United States -- primarily Meta's LLaMA -- has fueled deeper questions about the country's innovation model, which in recent decades surprised many by turning out world-beating firms like Alibaba and ByteDance despite Beijing's authoritarian controls.
The situation is emblematic of a reality that many in China openly admit. Even as the country races to build generative A.I., Chinese companies are relying almost entirely on underlying systems from the United States. China now lags the United States in generative A.I. by at least a year and may be falling further behind, according to more than a dozen tech industry insiders and leading engineers, setting the stage for a new phase in the cutthroat technological competition between the two nations that some have likened to a cold war. "Chinese companies are under tremendous pressure to keep abreast of U.S. innovations," said Chris Nicholson, an investor with the venture capital firm Page One Ventures who focuses on A.I. technologies. The release of ChatGPT was "yet another Sputnik moment that China felt it had to respond to."
Jenny Xiao, a partner at Leonis Capital, an investment firm that focuses on A.I.-powered companies, said the A.I. models that Chinese companies build from scratch "aren't very good," leading to many Chinese firms often using "fine-tuned versions of Western models." She estimated China was two to three years behind the United States in generative A.I. developments. The jockeying for A.I. primacy has huge implications. Breakthroughs in generative A.I. could tip the global technological balance of power, increasing people's productivity, aiding industries and leading to future innovations, even as nations struggle with the technology's risks. As Chinese firms aim to catch up by turning to open-source A.I. models from the United States, Washington is in a difficult spot. Even as the United States has tried to slow China's advancements by limiting the sale of microchips and curbing investments, it has not held back the practice of openly releasing software to encourage its adoption. For China, the newfound reliance on A.I. systems from the United States -- primarily Meta's LLaMA -- has fueled deeper questions about the country's innovation model, which in recent decades surprised many by turning out world-beating firms like Alibaba and ByteDance despite Beijing's authoritarian controls.
“RALPH” The Wonder Llama (Score:2)
Assisted by
Earl J. Llama
Mike Q. Llama III
Sy Llama
Merle Z. Llama IX
Sounds like a soft propaganda piece. (Score:3)
"We're ahead of the curve, and if you try to regulate us everyone else will catch up and that would be bad. You don't want China to catch up with us, do you? Better let us do what we want."
Re: (Score:3)
Re: (Score:2)
Re: (Score:3)
The last part
Regulation doesn't stop things from happening. It just stops things from happening here.
For things like pollution, that makes sense: Shift all the gunk somewhere else.
For things like AI, it makes no sense. Skynet will still happen, but Xi Jinping will control it.
Re: (Score:2)
Sorry, wrong country. It's North Korea that built and tested the Type O Dong.
Alibaba and ByteDance were before (Score:2, Interesting)
Alibaba and ByteDance were an older story from before the new crackdown by Pooh Bear. The great fact is that the traitor investors, who ran off to China to try to undermine the workers of the West are now getting arrested in China. Partly just because that's what China does to independent rich people and to foreigners, but mostly because, surprise surprise, it turns out the traitors are corrupt as hell.
What's New? (Score:5, Interesting)
USA has a copy-gap (Score:2)
The US has a "copy gap" that must be closed! We have to learn to swipe foreign technology faster or fall behind faster copiers. Remember that what it does is more important than how you got it.
Re: What's New? (Score:2)
Re: (Score:2)
Obviously, China will get ahead. Like they already have done in many areas, including EV batteries and telecoms.
We make the same mistake over and over again. Assume if we just keep everything secret, or choke off the supply of hardware, China will never catch up. After all, they are a 3rd world commie shithole, with joke universities, right?
In reality, they win because they have a lot of talent, and a lot of money to throw at R&D. The US has lots of private investors too, but they chase fads and aren't
Re: (Score:2)
China's good at copying goods, processes, and technology that the US and the rest of the world originally invented. This has been true across multiple industries over the last 25 years as US and the rest of the world moved manufacturing to China. That move was only because China had lots and lots of cheap labor. It wasn't because China were smarter or somehow better at making things than anyone else.
Exactly. It's the old story of inexpensive labor.
And eventually the labor becomes more expensive, and a new crop of inexpensive labor is found.
And there are as always, two sides to that issue. We all know the "They're takin our jerbs" lament. And to some extent, that is true. But what of the people with these new jobs that might be low paying by our standards, but a good job, as an alternative to subsistence living?
The upside of this job churn is that eventually most people on earth are living lives
Could they even do it? (Score:3)
it has not held back the practice of openly releasing software to encourage its adoption
Could the US gov, as large as it is, even be able to outlaw FOSS?
I suspect they would try but it would end up like the alcohol thing.
Re: (Score:3)
You'd likely run afoul of the first amendment. They might rule that some sections of the GPL or any other license might be unenforceable, but just saying "you can't release the source for your code" - no, I don't think that could be done under the constitution.
Re: (Score:2)
They can and will do what they like, there is no punishment for violating the Constitution but there sure is for pissing off the government.
Re: (Score:3)
Well, things like munitions can be restricted for export....if they classify AI and other things as munitions they could try to hinder things.
I know they tried that with cryptography in the early 90's.
Didn't work so well, but...it is an arrow in th
Re: (Score:2)
They tried that with cryptography about 25 years ago. What happened is that they printed out the source code on paper, mailed it to Europe, and over there they scanned it and recompiled it. That was 1st amendment protected free speech and they couldn't block that.
Also, American cryptography algorithms weren't necessarily any better than for example Russian ones.
Over here, I would find a "not USA" directory in many FTP sites where all the cryptography stuff was stored.
Re:Could they even do it? (Score:5, Informative)
Could the US gov, as large as it is, even be able to outlaw FOSS?
America tried to control encryption software in the 1990s.
The main result was to shift the development of encryption software out of America, the exact opposite of what was intended.
Export of cryptography from the United States [wikipedia.org]
Re: Could they even do it? (Score:2)
Re:Could they even do it? NO (Score:2)
Re: (Score:2)
"It would end up like the crypto thing [wikipedia.org]," is a better analogy. Encryption was classified as a "munition" until 1996, and restricted for export. Technically there are still export controls for new encryption schemes today, but it's a paper tiger at this point. It's likely that any enforcement attempts would be successfully met with free speech challenges, as with DeCSS (although who knows what the SCOTUS would decide on anything these days).
China is stuck (Score:3)
Not introducing AI means their economy slides compared to the West. Introducing AI could put millions of Chinese out of work and give those out-of-work Chinese time and an axe to grind about they should keep the CCP leech alive.
China's been working on AGI since 2017 (Score:4, Informative)
So if you think LLM's are not the way to AGI, then China is probably ahead of the US.
https://cset.georgetown.edu/pu... [georgetown.edu]
Re: (Score:2)
I do wonder, if AGI is created, and it's goal was betterment of people, will AGI actually be in favour of CCP's rule in China?
Or will it be a true AGI if it's hobbled someway so that it always agrees with CCP?
Re: (Score:2)
So what? Researchers have been working on AGI longer than I've been alive. That researchers in China are working on it is not even slightly close to being news.
Jobs? (Score:3)
Re: (Score:1)
So, in a country of a billion or so people, why is a country like China desperate to develop tech that will kill a lot of middle class jobs? They are the factory of the world - do they also need to master AI as well?
Maybe they notice thst "tech that kills jobs" is the route by which the West surpassed them and became so wealthy and powerful. I mean, I know that's so obvious that they couldn't possibly miss it but some people seem to.
What does "behind" even mean? (Score:3)
The barrier to entry for AI (generative or otherwise) is incredibly low, and mostly consists of harvesting large quantities of training data (which China should excel at, TBH) as well as access to sufficient processing power to process that data. OpenAI doesn't have any magical insight into how a trained model will behave compared to anyone else in the field -- anyone with money could be up to speed in a month or two, tops. The algorithms themselves are well-established at this point, and the "secret sauce" is not very secret either: increase the resolution of your data and add more processing power.
I suspect the reason we don't see the field flooded with more startups (if you don't consider it saturated anyway) is because the business case is just not there. Let OpenAI and others take the risk that there's no pot of gold at the end of the rainbow. If an opportunity presents itself, it would be easy enough to start competing; at least no more difficult than starting today, but certainly cheaper to wait.